Unlocking In-Context Learning for Natural Datasets Beyond Language Modelling
Computer Vision Group, University of Freiburg（弗赖堡大学计算机视觉组）；Neuromedical A.I. Lab, Medical Center -- University of Freiburg（弗赖堡大学医学神经AI实验室）；Medical Physics, Medical Center -- University of Freiburg（弗赖堡大学医学物理）；Visual Geometry Group, University of Oxford（牛津大学视觉几何组）
Jelena Bratulić, Sudhanshu Mittal, David T. Hoffmann, Samuel Böhm, Robin Tibor Schirrmeister, Tonio Ball, Christian Rupprecht, Thomas Brox
备注：
Best Paper Honorable Mention at GCPR 2025 (German Conference on Pattern Recognition). This is the updated version submitted to the conference, not the official conference proceedings
Large Language Models (LLMs) exhibit In-Context Learning (ICL), which enables the model to perform new tasks conditioning only on the examples provided in the context without updating the model's weights. While ICL offers fast adaptation across natural language tasks and domains, its emergence is less straightforward for modalities beyond text. In this work, we systematically uncover properties present in LLMs that support the emergence of ICL for autoregressive models and various modalities by promoting the learning of the needed mechanisms for ICL. We identify exact token repetitions in the training data sequences as an important factor for ICL. Such repetitions further improve stability and reduce transiency in ICL performance. Moreover, we emphasise the significance of training task difficulty for the emergence of ICL. Finally, by applying our novel insights on ICL emergence, we unlock ICL capabilities for various visual datasets and a more challenging EEG classification task.
CLUTCH: Contextualized Language model for Unlocking Text-Conditioned Hand motion modelling in the wild
Technical University of Darmstadt（德累斯顿技术大学）；Max-Planck Institute for Intelligent Systems（马克斯·普朗克智能系统研究所）；University of Tuebingen（图宾根大学）；Tubingen AI Center（图宾根人工智能中心）；Max Planck Institute for Informatics（马克斯·普朗克信息研究所）
Beyond the Covariance Trap: Unlocking Generalization in Same-Subject Knowledge Editing for Large Language Models
Institute of Information Engineering, Chinese Academy of Sciences（中国科学院信息工程研究所）；School of Cyber Security, University of Chinese Academy of Sciences（中国科学院大学网络安全学院）
Beyond Statistical Co-occurrence: Unlocking Intrinsic Semantics for Tabular Data Clustering
Department of Computer Science, Hong Kong Baptist University（香港浸会大学计算机科学系）；Department of Computer Science, Guangdong University of Technology（广东工业大学计算机科学系）
SOUL: Unlocking the Power of Second-Order Optimization for LLM Unlearning
Michigan State University（密歇根州立大学）；Lawrence Livermore National Laboratory（劳伦斯利弗莫尔国家实验室）；MIT-IBM Watson AI Lab, IBM Research（麻省理工学院-IBM沃森人工智能实验室，IBM研究院）
DreamGen: Unlocking Generalization in Robot Learning through Video World Models
NVIDIA；University of Washington（华盛顿大学）；KAIST（韩国科学技术院）；UCLA（加州大学洛杉矶分校）；UCSD（加州大学圣地亚哥分校）；CalTech（加州理工学院）；NTU（国立台湾大学）；University of Maryland（马里兰大学）；UT Austin（得克萨斯大学奥斯汀分校）
PRISM2: Unlocking Multi-Modal General Pathology AI with Clinical Dialogue
Paige, NYC, NY United States（美国纽约市帕伊公司）；Microsoft Research, Cambridge, MA United States（微软研究院）；Memorial Sloan Kettering Cancer Center, NYC, NY United States（纪念斯隆凯特林癌症中心）；University of Yale, New Haven, CT United States（耶鲁大学）
Beyond Reasoning: Reinforcement Learning Unlocks Parametric Knowledge in LLMs
State Key Laboratory of AI Safety, Institute of Computing Technology, CAS University of Chinese Academy of Sciences（人工智能安全国家重点实验室，计算技术研究所，中国科学院，中国科学院大学）
